Candlestick Patterns: 10 to Know, Tested on Real Data

Most candlestick pattern guides show you the perfect example and stop there. This one also shows you how often each pattern turned up when we counted it with fixed rules, and what the market did in the days after. The short version: some patterns appear every week or two, others only a few times in more than a decade, and for most of them we found too few cases to compare the next five sessions with similar days at all.

I have traded price since 1980, and candlesticks have been part of my daily work for decades. Here you get ten patterns, grouped by the number of candles they need. For each one I explain what it looks like, what it says about buyers and sellers, how traders usually handle it and what our count of the DAX, the S&P 500, EUR/USD and Gold since 2014 showed. If single candles are still new to you, read the guide on how to read candlestick charts first. This article follows our editorial policy.

Candlestick patterns in 30 seconds

• One to three candles: a candlestick pattern is a fixed arrangement of candles that traders read as a hint at a reversal or a continuation.
• Frequency varies widely: in almost 13 years of daily candles we found 283 to 318 dojis per market, but only 4 to 9 morning stars.
• Small samples, wide intervals: for doji and inside bar the 95% interval of the difference to similar days contained zero in all 16 comparisons; most patterns were too rare for any comparison.
• Location first: I act on a pattern only at support, at resistance or after a clear move, and only with a stop beyond the pattern.

What are candlestick patterns?

A candlestick pattern is a recognizable arrangement of one to three candles that shows a shift in the balance between buyers and sellers. Traders use these patterns to time entries and exits, usually as a hint at a reversal or a continuation. Some patterns are bullish, some bearish, and a few, like the doji, have no direction at all.

Candlestick patterns are much shorter than chart patterns. A head and shoulders top or a flag needs weeks of price data, while a candlestick pattern is complete after one to three periods. The Western standard texts are Steve Nison’s 1991 book on Japanese candlestick charting and Thomas Bulkowski’s Encyclopedia of Candlestick Charts with 103 patterns; both are discussed in my list of trading books.

The ten patterns: a cheat sheet

The selection of ten patterns goes back to Galen Woods. Galen Woods was a valued partner of Kagels Trading, and our German site published German translations of his articles. The German version of this page started as a translation of his overview 10 Price Action Candlestick Patterns You Must Know, which I recommend reading alongside this one. The grouping, the explanations, the counting rules, the measurement and the charts on this page are our own.

Pattern (bullish / bearish)	Candles	What it shows
Doji (neutral)	1	stalemate
Bullish / bearish marubozu	1	one side dominates
Hammer / hanging man	1	possible reversal
Inverted hammer / shooting star	1	possible reversal
Bullish / bearish engulfing	2	takeover
Bullish / bearish harami	2	move stalls
Piercing line / dark cloud cover	2	possible reversal
Morning star / evening star	3	reversal
Three white soldiers / three black crows	3	strong move
Bullish / bearish hikkake	3, plus up to 3 for confirmation	false breakout

To count patterns, you need rules that leave no room for interpretation. The table gives our rules in short. Size thresholds refer to the ATR of the candle before the pattern, and all other comparisons are exact. Our full results table adds the remaining patterns we measured, such as tweezers and pin bars.

Pattern	Our counting rule in short
Doji	body at most 10% of the range
Marubozu	body at least 90% of the range, range at least 1 ATR
Hammer, hanging man	body above 10% of the range, lower wick at least twice the body, upper wick at most 10%
Inverted hammer, shooting star	body above 10% of the range, upper wick at least twice the body, lower wick at most 10%
Engulfing	second body covers the first and is larger
Harami	first body at least 0.6 ATR, second inside it and at most half its size
Piercing line, dark cloud cover	first body at least 0.6 ATR, open beyond the prior close, close beyond the middle of the first body
Morning star, evening star	first body at least 0.6 ATR, small middle candle near the first close, third closes beyond the middle of the first, no gap required
Three soldiers, three crows	three strong candles, each opening inside the prior body

Most patterns also need the right market state before them. We call the market state down when the close is below the close of ten sessions earlier and below the 20-day average, and up in the opposite case. Hammer, inverted hammer, bullish engulfing, harami, piercing line and morning star need a down state; their bearish counterparts need an up state. Three white soldiers may also follow a sideways market, three black crows as well, while the marubozu, the doji and the inside bar have no trend condition.

What our count showed

We applied these rules to daily candles of four markets. The data are the DAX and the S&P 500 from Yahoo Finance and EUR/USD and Gold from Dukascopy, with the day ending at 5 pm New York, from January 2014 to September 2026. That is between 3,205 and 3,318 sessions per market. Only evaluable cases count: days with a data gap nearby and signals without ten complete following sessions drop out.

The spread between common and rare patterns is large. The inside bar and the doji appear several hundred times per market, the hammer 9 to 31 times and the morning star 4 to 9 times. We did not find one three white soldiers in the DAX. The example charts further down always show the most recent evaluable case of a pattern up to September 2026, picked without looking at the result.

Next we compared each pattern with similar days without it. For directional patterns we measured the move from the next open to the close five sessions later, in the direction of the pattern, in multiples of the average true range over 14 sessions. The comparison group consists of days from the same year, with the same market state and a similar volatility level. We fixed the rules before the evaluation, developed them on 2014 to 2021 and kept 2022 to 2026 as the test period. We show a value only where at least 30 comparable cases existed, so most patterns, markets and periods are missing from the table.

Pattern, market, period	Comparable cases	Difference in ATR
Bearish engulfing, DAX, 2014 to 2021	30 of 30	+0.46 (−0.06 to +0.89)
Bearish engulfing, S&P 500, 2014 to 2021	45 of 45	−0.43 (−0.88 to +0.03)
Bearish engulfing, EUR/USD, 2014 to 2021	41 of 41	−0.10 (−0.54 to +0.44)
Bearish engulfing, Gold, 2014 to 2021	62 of 62	−0.11 (−0.55 to +0.25)
Bearish engulfing, Gold, 2022 to 2026	36 of 36	−0.23 (−0.86 to +0.32)
Bearish harami, EUR/USD, 2014 to 2021	30 of 30	+0.34 (−0.01 to +0.72)
Bearish marubozu, DAX, 2014 to 2021	33 of 38	+0.19 (−0.41 to +0.83)
Bearish marubozu, S&P 500, 2014 to 2021	42 of 46	−0.45 (−1.07 to +0.10)
Bearish pin bar, EUR/USD, 2014 to 2021	44 of 45	+0.21 (−0.17 to +0.53)
Bearish pin bar, Gold, 2014 to 2021	32 of 35	−0.19 (−0.77 to +0.35)
Bullish engulfing, EUR/USD, 2014 to 2021	84 of 84	+0.07 (−0.27 to +0.36)
Bullish engulfing, EUR/USD, 2022 to 2026	34 of 34	−0.45 (−1.02 to +0.17)
Bullish engulfing, Gold, 2014 to 2021	53 of 53	+0.38 (+0.02 to +0.70)
Bullish marubozu, S&P 500, 2014 to 2021	50 of 52	−0.10 (−0.63 to +0.31)
Bullish pin bar, S&P 500, 2014 to 2021	32 of 35	−0.34 (−1.27 to +0.36)
Hanging man, DAX, 2014 to 2021	31 of 31	−0.33 (−0.89 to +0.28)
Hanging man, S&P 500, 2014 to 2021	46 of 46	−0.08 (−0.45 to +0.34)
Tweezer bottom, EUR/USD, 2014 to 2021	34 of 34	+0.20 (−0.26 to +0.65)
Tweezer top, S&P 500, 2014 to 2021	43 of 43	+0.03 (−0.30 to +0.43)
Tweezer top, Gold, 2014 to 2021	34 of 34	−0.21 (−0.88 to +0.45)

Read the table as a description, not as a ranking. The brackets hold the 95% interval from a block bootstrap. In 19 of the 20 rows it includes zero; for the bullish engulfing in Gold from 2014 to 2021 it runs from +0.02 to +0.70. For doji and inside bar we compared the range of the next five sessions instead of a direction; the differences lay between −0.13 and +0.28 ATR, and all 16 intervals included zero. With this many comparisons across patterns and markets, a few striking values are expected by chance alone.

What the count does not tell you

The count describes four markets on daily candles, nothing more. It covers no other time frames and no other markets, it ignores costs, and it does not test any of the trading approaches described below. For directional patterns we measured the move from the next open to the close five sessions later, for doji and inside bar the range of those five sessions. An interval that includes zero does not prove that a pattern is useless, and a pattern with fewer than 30 comparable cases was not compared with similar days at all. Index levels cannot be traded directly, and EUR/USD and Gold are bid prices.

Some of our rules deviate from common textbook versions. For the piercing line, textbooks often require an open below the prior low, and for the dark cloud cover an open above the prior high; we require only an open beyond the prior close. Morning and evening stars are counted without a gap, because EUR/USD and Gold trade around the clock. In Gold the opening price depends on the data source, and the minimum coverage of a Gold session was set to 18 hours after the case counts were known, so the Gold results are descriptive and not an untouched out-of-sample check.

Everything is available for your own checks. The CSV linked above lists every pattern, market and period of protocol version 3.2, including the many rows below 30 comparable cases that show only the case count, and the intervals for block lengths of 10, 20 and 40 sessions. Belt hold, hikkake, abandoned baby and the gap versions of the star patterns are not part of the measurement.

Single-candle patterns

One candle can already tell you which side controlled a period. The four single-candle patterns differ in two things: the size of the body and the position of the wicks. They are the quickest to spot and the easiest to misread, because one candle is a very small sample of market behavior.

Doji

A doji is a candle whose open and close are almost the same, so its body shrinks to a thin line. Neither buyers nor sellers won the period. We count a candle as a doji when its body is at most 10% of its range. The wicks give the variants their names: the long-legged doji has long wicks on both sides, the dragonfly doji a long lower wick, the gravestone doji a long upper wick.

On its own a doji is ordinary. With 283 to 318 cases per market it appeared about once every ten to twelve sessions in our data. Inside a sideways range it adds little; after a long rise into resistance many traders take it as a warning that buying pressure is fading. That reading depends on the location, not on the candle.

Traders who act on a doji usually wait for the next candle. After a doji at resistance, a break of the doji low is the common trigger for a short trade, with the stop above the doji high. Without that break the doji stays an observation. We have not tested this approach.

Our most recent DAX doji formed on September 16, 2026. In the five sessions after it the DAX moved within a range of 2.02 times its ATR. Across all eight doji comparisons the interval of the difference to similar days included zero, so our data do not show that a doji was followed by more movement than usual.

Marubozu

A marubozu is a candle that consists almost only of body. The open sits near one end of the range, the close near the other, and the wicks are tiny or missing. A bullish marubozu shows buyers in control from open to close, a bearish marubozu sellers.

Our rule asks for more than a large body. The body must fill at least 90% of the range, and the candle must be at least one ATR long, so that a quiet day with a small range does not qualify. Counted that way, bullish marubozus appeared 43 times in the DAX, 67 in the S&P 500, 34 in EUR/USD and 44 in Gold.

A close relative is the belt hold, which opens at its low and closes near its high but may keep an upper wick. Galen Woods describes two ways to trade the belt hold in a separate article. We have not included the belt hold in our count, so the numbers on this page do not cover it, and our rules treat it as a separate pattern from the marubozu.

After a marubozu the move is often already well under way. That is why many traders use it as a sign of direction rather than as an entry. One approach waits for a pullback into the lower half of a bullish body and places the stop below the candle’s low; a gap in the same direction the next day leaves this approach without an entry. We have not tested it.

Our most recent bullish marubozu in Gold formed on August 19, 2026. Five sessions later Gold stood +0.67 ATR higher than at the next open. One case shows what the pattern looks like on a real chart; it says nothing about how such cases usually develop.

Hammer and hanging man

Hammer and hanging man are the same candle in two different places. Both have a small body near the top of the range, a long lower wick and almost no upper wick. After a decline the candle is called a hammer, after a rise a hanging man.

A hammer shows that sellers pushed price down and buyers pushed it back up before the close. In our rule the lower wick is at least twice the body, the upper wick at most 10% of the range, and the color does not matter. The pin bar is a close relative whose wick also reaches below the lows of the prior five candles; in our data hammers and pin bars often fall on the same candle.

The usual entry after a hammer lies above its high, with the stop below the end of the long wick. Because that wick is long, the risk per trade is often large, so I derive the position size from the stop distance and never the other way round; how that works is explained in the guide on position sizing. For the hanging man all levels are mirrored.

Our most recent DAX hammer formed on July 16, 2026, and five sessions later the DAX stood +0.02 ATR above the next open. We counted 31 hammers in the DAX, 19 in the S&P 500, 17 in EUR/USD and 9 in Gold. The hanging man in the DAX from 2014 to 2021 was frequent enough for a comparison: −0.33 ATR against similar days, with an interval from −0.89 to +0.28.

Inverted hammer and shooting star

Turn a hammer upside down and you get the inverted hammer and the shooting star. The long wick points up and the small body sits near the low of the range. After a rise the candle is a shooting star, after a decline an inverted hammer.

A shooting star records a rally that did not hold until the close. Our rule requires an upper wick of at least twice the body and a lower wick of at most 10% of the range; unlike some textbooks, we do not require a gap to the prior candle. Shooting stars appeared 26 times in the DAX, 21 in the S&P 500, 34 in EUR/USD and 29 in Gold.

A common way to trade the shooting star waits for a break of its low. The stop goes above the tip of the wick. Near a resistance zone the candle has a reason to matter; far from any level it is just a candle with a long wick, as I explain in the guide on support and resistance.

Our most recent shooting star in Gold formed on December 5, 2025. Five sessions later Gold stood −1.32 ATR in the direction of the pattern, which means it had risen against the bearish reading. Such a single case shows the range of possible outcomes, not their frequency.

Two-candle patterns

Two-candle patterns compare one period with the period before it. The question is always the same: how far did the second candle take back what the first one did? The answers range from a complete takeover to a pause, and all three patterns in this group need a prior move to have something to reverse.

Engulfing

In an engulfing pattern the body of the second candle covers the body of the first completely. In a bullish engulfing after a decline, a green candle opens at or below the prior red close and closes at or above the prior red open. The bearish engulfing after a rise is the mirror image. If the second candle also covers the wicks of the first, we count it as a separate whole-candle variant.

For me an engulfing is worth a look only at a zone the market has respected before. With the bearish version the stop goes above the high of both candles and the first target at the next support below. If that target is closer than the stop, the trade does not pay for its risk, and I leave it.

Our most recent bearish engulfing in Gold formed on September 1, 2026. Five sessions later Gold stood −0.27 ATR in the direction of the pattern, so it had moved against it. The bearish engulfing was one of the more frequent patterns, with 54 cases in the DAX, 69 in the S&P 500, 60 in EUR/USD and 98 in Gold. From 2014 to 2021 the difference to similar days was +0.46 ATR in the DAX and −0.43 ATR in the S&P 500, both with intervals that include zero.

Harami

A harami is the opposite of an engulfing: the second body sits entirely inside the first. A large candle in the direction of the move is followed by a small one, a sign that the move has paused. The word is Japanese for pregnant, a picture of the small candle inside the large one. If the small candle is a doji, the pattern is a harami cross.

Our rule makes the first candle large and the second clearly small. The first body must be at least 0.6 ATR, and the second body at most half its size. The inside bar is related but stricter in one way and looser in another: there the whole range of the second candle, wicks included, stays inside the first range, with no condition on the bodies.

A harami needs a break before it becomes a trade. A common rule for the bullish harami puts the entry above the high of the large red candle and the stop below its low. Without that break the harami is only a pause inside the downtrend.

Our most recent bullish harami in Gold ended on September 11, 2026. Five sessions later Gold stood +0.41 ATR above the next open. We counted 28 bullish haramis in the DAX, 15 in the S&P 500, 32 in EUR/USD and 29 in Gold, too few for a comparison in any market or period.

Piercing line and dark cloud cover

Piercing line and dark cloud cover are half-way versions of the engulfing. The second candle reaches past the middle of the first body but not beyond its open. The counter-move is visible, but it does not complete the takeover.

The piercing line comes after a large red candle in a decline. A green candle opens below the red close and closes above the middle of the red body. The dark cloud cover is the same idea after a rise: a red candle opens above a large green candle’s close and closes below the middle of its body.

Traders often treat both like a weaker engulfing and wait for confirmation. For the piercing line that means an entry only above its high and a stop below the low of both candles. In our data both were rare: 4 to 26 piercing lines and 6 to 18 dark cloud covers per market, too few for a comparison.

Three-candle patterns

Three-candle patterns tell a short story: a move, a hesitation or a continuation, and a result. Because they need three specific candles in a row, they are the rarest patterns in our count. That rarity alone is a reason to be careful with promises about them and with conclusions drawn from single charts.

Morning star and evening star

A morning star is a large red candle, a small candle and a large green candle that closes above the middle of the first. The small middle candle shows hesitation after the decline, the third candle the counter-move. The evening star is the mirror image after a rise. The name refers to the morning star that appears before sunrise.

Many textbook drawings separate the middle candle from the other two with gaps. We count without that gap, because EUR/USD and Gold trade around the clock and such gaps hardly occur there, and we mark the pattern “without gap” in charts and captions. Traders who act on a morning star usually place the stop a little below the lowest low of the three candles, and for the evening star above the highest high.

Both patterns were rare in our data. We found 4 to 9 morning stars and 2 to 8 evening stars per market in almost 13 years, too few for a comparison with similar days. The two DAX examples below happen to point in opposite directions.

Our most recent DAX morning star ended on June 12, 2026, and five sessions later the DAX stood −0.22 ATR in the direction of the pattern. The most recent evening star in the DAX, ending on March 2, 2026, was followed by a decline: +2.66 ATR in the direction of the pattern. Two examples, two outcomes: looking at one chart tells you what a pattern can look like, not how it usually behaves.

Three white soldiers and three black crows

Three white soldiers are three strong green candles in a row, each opening inside the prior body and closing higher. Three black crows are three strong red candles, each closing lower. In our rule each body is at least half an ATR, and the wicks on the side of the move stay small.

Under these rules both patterns almost never appeared. We found three white soldiers 0 times in the DAX, once in the S&P 500, 6 times in EUR/USD and twice in Gold, and three black crows 1, 2, 1 and 3 times. That is too few for any statistic, and a reminder that a pattern can be famous and still rare.

Buying after three strong days means buying after the market has already moved. A more cautious approach waits for the first pullback and treats it like any pullback in the direction of the move. We have not tested this approach.

The hikkake: a pattern built on a failed breakout

The hikkake is a trap for breakout traders. The analyst Dan Chesler described it: an inside bar forms, price breaks out of it, the breakout fails, and price runs the other way.

The setup needs three bars, and the confirmation can take up to three more. First a mother bar, then an inside bar within its range, then a trap bar that breaks out of the inside bar: for the bullish hikkake a bar with a lower high and a lower low. The pattern is confirmed when price trades above the inside bar’s high, usually within three bars after the trap. The whole sequence therefore spans four to six bars, which is one reason we did not include the hikkake in our count.

The entry lies at the confirmation, above the high of the inside bar for the bullish version. The stop goes below the low of the trap bar, which often makes the risk small. One possible explanation for the move: traders who bought or sold the false breakout close their positions at a loss. We have not tested the pattern, and how I handle breakouts in general is a separate topic.

From pattern to trade plan in four steps

A pattern gives me the timing, the location gives me the reason. That is the core of my approach to price action trading, and I work without technical indicators. These four steps turn a candle shape into a plan:

1. Location: is the pattern at support, at resistance or at a Fibonacci zone? If not, there is no trade.
2. Trend: do the highs and lows form a rising or a falling sequence? I read this from the market structure; traders who use indicators often take a moving average.
3. Confirmation: I enter only when price breaks the high or the low of the pattern.
4. Stop and size: the stop goes beyond the pattern’s extreme, and the position size follows from that distance.

The exit belongs in the plan before the entry. A natural first target is the next zone in the direction of the trade. Record every trade and review your notes regularly, so you can see over time whether your rules are working; the tools I compared for that are in the review of trading journal software. How much to risk per trade is covered in the guide on risk management.

The time frame changes what a pattern means. On a daily chart one candle covers a whole trading day, on a five-minute chart five minutes. For swing trades I prefer daily candles. On short time frames patterns appear more often, and single candles are more exposed to random moves; our count covers daily candles only.

Smoothed candles do not work for these patterns. Heikin Ashi charts recalculate each candle from averages, so they do not show the real open and close. For candlestick patterns you need a normal candlestick chart, which every common platform provides.

Candlestick analysis in eight questions

Galen Woods developed a checklist that puts every candle into context. He published it as Eight Questions to Help You with Price Action Bar Analysis. I summarize his questions here in my own words, because they move your attention from the pattern to its surroundings:

1. Wicks: how long are the upper and lower wicks, and where were prices rejected?
2. Trend bar: does the candle have a large body that clearly shows one side in front?
3. Sequence: do several candles run in the same direction one after another?
4. Overlap: how much does the candle overlap with the one before?
5. Gap: how did the market react to a gap at the open?
6. Test of a zone: what happened when price reached support or resistance?
7. Three candles: what do the last three candles say together?
8. Twenty candles: in which environment do the last twenty candles stand?

In my experience the last two questions matter most. They force you to step back from the single candle to the whole picture. That habit separates a trader who collects patterns from one who reads markets.

Common mistakes with candlestick patterns

• Ignoring the location: a hammer in the middle of a move is a candle with a long wick, nothing more.
• Acting before the close: a candle is finished only at its close. Until then a hammer can still turn into a large red body.
• Accepting wide stops without adjusting size: long wicks mean wide stops; if the position size stays the same, the risk per trade grows.
• Waiting for rare patterns: under strict rules, a morning star comes down to a handful of cases per market in more than twelve years. A few common patterns with clear rules are more practical.
• Taking one chart as proof: a good-looking example proves nothing either way. Only many cases show how a pattern usually behaves.

Conclusion: know the shapes, judge the location

The ten patterns on this page belong to the basic vocabulary of anyone who reads candlestick charts. You should recognize them quickly. Our count adds two facts: many of them are rare under clear rules, and where we had enough cases on daily candles of four markets, the interval of the difference to similar days included zero in 35 of 36 comparisons.

What turns a candle shape into a trade is the plan around it. A pattern at a tested zone, an entry only after the break and a stop beyond the extreme: with those three elements a pattern becomes a trade plan you can review. The rest is practice and an honest journal.

Frequently asked questions about candlestick patterns

Which candlestick patterns should you know?

Ten patterns cover most of what you will meet. They are the doji, the marubozu, the hammer and hanging man, the inverted hammer and shooting star, the engulfing, the harami, the piercing line and dark cloud cover, the morning and evening star, the three soldiers and crows, and the hikkake.

Do candlestick patterns work?

Our count measured what followed the patterns, not whether trading them works. Where at least 30 comparable cases existed, the 95% interval of the difference to similar days included zero in 19 of 20 directional comparisons; for the bullish engulfing in Gold from 2014 to 2021 it ran from +0.02 to +0.70 ATR. For doji and inside bar, measured by the range of the next five sessions, all 16 intervals included zero. Most patterns had too few cases for a comparison, and the trading approaches on this page were not tested, so these data describe the past in four markets and are no verdict on any strategy.

How many candlestick patterns are there?

More than a hundred candlestick patterns have been described. Thomas Bulkowski lists 103 in his encyclopedia. Most traders use only a small part of them, and the ten on this page cover the common cases.

Which time frame is best for candlestick patterns?

The daily chart is a good place to start. Each candle stands for a whole trading day, so single random moves weigh less than on short time frames. On a five-minute chart patterns appear far more often, and our count covers daily candles only.

What are bullish and bearish candlestick patterns?

Bullish patterns hint at rising prices after a decline, bearish patterns at falling prices after a rise. Bullish examples are the hammer, the bullish engulfing and the morning star; bearish examples are the shooting star, the bearish engulfing and the evening star. The doji is neutral.

Is there a candlestick patterns cheat sheet?

The cheat sheet near the top of this page lists all ten patterns with their bullish and bearish versions. Below it you find the short rule we used to count each pattern, and the CSV with all results for your own analysis.

What is the difference between a candlestick pattern and a chart pattern?

The difference is the number of candles. A candlestick pattern usually consists of one to three candles. A chart pattern such as head and shoulders or a flag forms over many candles and often over weeks.